From be2daab00cabc526e12754d749723fbf98b705b9 Mon Sep 17 00:00:00 2001
From: Abdessamad Derraz <3028866+Abdess@users.noreply.github.com>
Date: Fri, 9 Oct 2026 22:45:57 +0200
Subject: [PATCH] fix: judge the gap report on the flat declared set

---
 scripts/cross_reference.py   | 30 +++++++++++++++++++---
 tests/test_declared_names.py | 49 ++++++++++++++++++++++++++++++++++++
 2 files changed, 75 insertions(+), 4 deletions(-)
 create mode 100644 tests/test_declared_names.py

diff --git a/scripts/cross_reference.py b/scripts/cross_reference.py
index 02be2fd8..488ad8e4 100644
--- a/scripts/cross_reference.py
+++ b/scripts/cross_reference.py
@@ -70,6 +70,25 @@ def load_platform_files(
     return declared, platform_data_dirs
 
 
+def declared_names(platforms_dir: str, platforms: list[str] | None, db: dict) -> set[str]:
+    """Every name the platforms declare, under every spelling the database knows.
+
+    The report judged "undeclared" per normalized system id, which does not
+    bring bk and elektronika-bk, or nintendo-sgb and nintendo-super-game-boy,
+    together: 646 entries whose exact name a platform declares were counted
+    undeclared, while verify and the site read one flat set.
+    """
+    from common import expand_platform_declared_names
+
+    names: set[str] = set()
+    for platform_name in platforms or list_registered_platforms(
+        platforms_dir, include_archived=True
+    ):
+        config = load_platform_config(platform_name, platforms_dir)
+        names.update(expand_platform_declared_names(config, db))
+    return names
+
+
 def _build_supplemental_index(
     data_root: str = "data", bios_root: str = "bios"
 ) -> set[str]:
@@ -135,6 +154,7 @@ def _resolve_source(
     by_path_suffix: dict | None = None,
     file_entry: dict | None = None,
     db_files: dict | None = None,
+    bios_dir: str | None = None,
 ) -> str | None:
     """Return the source category for a file, or None if not found.
 
@@ -177,7 +197,7 @@ def _resolve_source(
         return "bios"
     # bios/ under the canonical MAME set name, as resolve_local_file does:
     # a renamed archive is held once, under the name the dedup kept.
-    canonical = get_mame_clone_map().get(fname)
+    canonical = get_mame_clone_map(bios_dir).get(fname)
     if canonical and canonical != fname:
         if canonical in by_name and _name_hit(canonical):
             return "bios"
@@ -269,6 +289,7 @@ def entry_source(f: dict, index: dict) -> str | None:
             continue
         source = _resolve_source(
             candidate, by_name, by_name_lower, data_names, by_path_suffix, f, db_files,
+            index.get("bios_dir"),
         )
         if source is not None:
             return source
@@ -513,6 +534,7 @@ def cross_reference(
         "db_files": db_files,
         "data_names": data_names,
         "all_declared": all_declared,
+        "bios_dir": db.get("bios_dir"),
     }
     for emu_name, profile in profiles.items():
         _cross_reference_profile(
@@ -611,13 +633,13 @@ def main():
         print("No emulator profiles found.", file=sys.stderr)
         return
 
-    declared, plat_data_dirs = load_platform_files(
-        args.platforms_dir, [args.platform] if args.platform else None
-    )
+    selected = [args.platform] if args.platform else None
+    declared, plat_data_dirs = load_platform_files(args.platforms_dir, selected)
     db = load_database(args.db)
     data_names = _build_supplemental_index()
     report = cross_reference(
         profiles, declared, db, plat_data_dirs, data_names,
+        all_declared=declared_names(args.platforms_dir, selected, db),
         standalone_cores=standalone_cores,
     )
 
diff --git a/tests/test_declared_names.py b/tests/test_declared_names.py
new file mode 100644
index 00000000..6eeba15e
--- /dev/null
+++ b/tests/test_declared_names.py
@@ -0,0 +1,49 @@
+"""The command line report judges "undeclared" on the flat declared set.
+
+Per normalized system id, bk and elektronika-bk never met, nor nintendo-sgb
+and nintendo-super-game-boy: 646 entries whose exact name a platform
+declares were counted undeclared, while verify and the site read one flat
+set, enriched with every spelling the database knows.
+"""
+
+from __future__ import annotations
+
+import sys
+import unittest
+from pathlib import Path
+
+REPO_ROOT = Path(__file__).resolve().parent.parent
+sys.path.insert(0, str(REPO_ROOT / "scripts"))
+
+from common import load_database, load_emulator_profiles  # noqa: E402
+from cross_reference import (  # noqa: E402
+    _build_supplemental_index,
+    cross_reference,
+    declared_names,
+    load_platform_files,
+)
+
+
+class FlatDeclaredSet(unittest.TestCase):
+    def test_a_name_declared_under_another_system_id_is_not_a_gap(self):
+        if not (REPO_ROOT / "database.json").is_file():
+            self.skipTest("no database.json")
+        db = load_database(str(REPO_ROOT / "database.json"))
+        platforms = str(REPO_ROOT / "platforms")
+        profiles = {k: v for k, v in load_emulator_profiles(str(REPO_ROOT / "emulators")).items() if k == "bk"}
+        if not profiles:
+            self.skipTest("no bk profile")
+        declared, data_dirs = load_platform_files(platforms, ["batocera"])
+        names = _build_supplemental_index()
+        per_system = cross_reference(profiles, declared, db, data_dirs, names)
+        flat = cross_reference(
+            profiles, declared, db, data_dirs, names,
+            all_declared=declared_names(platforms, ["batocera"], db),
+        )
+        gaps = {g["name"] for g in flat["bk"]["gap_details"]}
+        self.assertNotIn("MONIT10.ROM", gaps)
+        self.assertLess(flat["bk"]["gaps"], per_system["bk"]["gaps"])
+
+
+if __name__ == "__main__":
+    unittest.main()